1 /* ----------------------------------------------------------------------
2 * Project: CMSIS DSP Library
3 * Title: arm_add_f32.c
4 * Description: Floating-point vector addition
5 *
6 * $Date: 23 April 2021
7 * $Revision: V1.9.0
8 *
9 * Target Processor: Cortex-M and Cortex-A cores
10 * -------------------------------------------------------------------- */
11 /*
12 * Copyright (C) 2010-2021 ARM Limited or its affiliates. All rights reserved.
13 *
14 * SPDX-License-Identifier: Apache-2.0
15 *
16 * Licensed under the Apache License, Version 2.0 (the License); you may
17 * not use this file except in compliance with the License.
18 * You may obtain a copy of the License at
19 *
20 * www.apache.org/licenses/LICENSE-2.0
21 *
22 * Unless required by applicable law or agreed to in writing, software
23 * distributed under the License is distributed on an AS IS BASIS, WITHOUT
24 * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
25 * See the License for the specific language governing permissions and
26 * limitations under the License.
27 */
28
29 #include "dsp/basic_math_functions.h"
30
31 /**
32 @ingroup groupMath
33 */
34
35 /**
36 @defgroup BasicAdd Vector Addition
37
38 Element-by-element addition of two vectors.
39
40 <pre>
41 pDst[n] = pSrcA[n] + pSrcB[n], 0 <= n < blockSize.
42 </pre>
43
44 There are separate functions for floating-point, Q7, Q15, and Q31 data types.
45 */
46
47 /**
48 @addtogroup BasicAdd
49 @{
50 */
51
52 /**
53 @brief Floating-point vector addition.
54 @param[in] pSrcA points to first input vector
55 @param[in] pSrcB points to second input vector
56 @param[out] pDst points to output vector
57 @param[in] blockSize number of samples in each vector
58 */
59
60 #if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
61
62 #include "arm_helium_utils.h"
63
arm_add_f32(const float32_t * pSrcA,const float32_t * pSrcB,float32_t * pDst,uint32_t blockSize)64 void arm_add_f32(
65 const float32_t * pSrcA,
66 const float32_t * pSrcB,
67 float32_t * pDst,
68 uint32_t blockSize)
69 {
70 uint32_t blkCnt; /* Loop counter */
71
72 f32x4_t vec1;
73 f32x4_t vec2;
74 f32x4_t res;
75
76 /* Compute 4 outputs at a time */
77 blkCnt = blockSize >> 2U;
78
79 while (blkCnt > 0U)
80 {
81 /* C = A + B */
82
83 /* Add and then store the results in the destination buffer. */
84 vec1 = vld1q(pSrcA);
85 vec2 = vld1q(pSrcB);
86 res = vaddq(vec1, vec2);
87 vst1q(pDst, res);
88
89 /* Increment pointers */
90 pSrcA += 4;
91 pSrcB += 4;
92 pDst += 4;
93
94 /* Decrement the loop counter */
95 blkCnt--;
96 }
97
98 /* Tail */
99 blkCnt = blockSize & 0x3;
100
101 if (blkCnt > 0U)
102 {
103 /* C = A + B */
104 mve_pred16_t p0 = vctp32q(blkCnt);
105 vec1 = vld1q(pSrcA);
106 vec2 = vld1q(pSrcB);
107 vstrwq_p(pDst, vaddq(vec1,vec2), p0);
108 }
109
110 }
111
112 #else
arm_add_f32(const float32_t * pSrcA,const float32_t * pSrcB,float32_t * pDst,uint32_t blockSize)113 void arm_add_f32(
114 const float32_t * pSrcA,
115 const float32_t * pSrcB,
116 float32_t * pDst,
117 uint32_t blockSize)
118 {
119 uint32_t blkCnt; /* Loop counter */
120
121 #if defined(ARM_MATH_NEON) && !defined(ARM_MATH_AUTOVECTORIZE)
122 f32x4_t vec1;
123 f32x4_t vec2;
124 f32x4_t res;
125
126 /* Compute 4 outputs at a time */
127 blkCnt = blockSize >> 2U;
128
129 while (blkCnt > 0U)
130 {
131 /* C = A + B */
132
133 /* Add and then store the results in the destination buffer. */
134 vec1 = vld1q_f32(pSrcA);
135 vec2 = vld1q_f32(pSrcB);
136 res = vaddq_f32(vec1, vec2);
137 vst1q_f32(pDst, res);
138
139 /* Increment pointers */
140 pSrcA += 4;
141 pSrcB += 4;
142 pDst += 4;
143
144 /* Decrement the loop counter */
145 blkCnt--;
146 }
147
148 /* Tail */
149 blkCnt = blockSize & 0x3;
150
151 #else
152 #if defined (ARM_MATH_LOOPUNROLL) && !defined(ARM_MATH_AUTOVECTORIZE)
153
154 /* Loop unrolling: Compute 4 outputs at a time */
155 blkCnt = blockSize >> 2U;
156
157 while (blkCnt > 0U)
158 {
159 /* C = A + B */
160
161 /* Add and store result in destination buffer. */
162 *pDst++ = (*pSrcA++) + (*pSrcB++);
163 *pDst++ = (*pSrcA++) + (*pSrcB++);
164 *pDst++ = (*pSrcA++) + (*pSrcB++);
165 *pDst++ = (*pSrcA++) + (*pSrcB++);
166
167 /* Decrement loop counter */
168 blkCnt--;
169 }
170
171 /* Loop unrolling: Compute remaining outputs */
172 blkCnt = blockSize % 0x4U;
173
174 #else
175
176 /* Initialize blkCnt with number of samples */
177 blkCnt = blockSize;
178
179 #endif /* #if defined (ARM_MATH_LOOPUNROLL) */
180 #endif /* #if defined(ARM_MATH_NEON) */
181
182 while (blkCnt > 0U)
183 {
184 /* C = A + B */
185
186 /* Add and store result in destination buffer. */
187 *pDst++ = (*pSrcA++) + (*pSrcB++);
188
189 /* Decrement loop counter */
190 blkCnt--;
191 }
192
193 }
194 #endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
195
196 /**
197 @} end of BasicAdd group
198 */
199
